[csglinux]

As @crinacle says, most likely due to ear canal resonance. The region around 7kHz is prone to that on my ears, too.

Try out different insertion depths (shallower / deeper). If the peak's frequency changes, it's almost certainly caused by your canal.

And/or your choice of eartips. They can make a huge difference:

https://www.head-fi.org/threads/shu...tatic-earphones.785377/page-207#post-13714398

I and @ohotonge also measured peaks around 10 kHz, but as others have mentioned, that will shift a bit with insertion depth.

I don't know what tips @crinacle used for his measurements, but it looks like Shure olives. With silicone tips, I'd definitely expect to see some peak up there around 10 kHz.

 

[bartzky]

Thanks both of you man.
It may be true. I have a question : does headphone manufacturers target to a response base on below chart (iso human hearing spl ) for their products?
And from the chart, most people have a peak at 3-4khz ?(with a true flat speaker)

No. At least I hope so.

What we're looking at are SPL levels for pure sine tones that will result in the same perceived loudness regardless of the frequency. Those graphs answer questions like: If 1 kHz plays at 80dB SPL, how loud must be 2 kHz to be recognized as equally loud?

Regarding IEM measurements we usually talk about about the frequency response at the eardrum (DRP). That's a whole nother story.

The only information we can take out of the above diagrams and use for IEM tuning is the difference between two of those curves.
Let's say your music is mastered to be heard at 80dB SPL (@1 kHz). But you want to protect your hearing and like to listen at 10dB less volume. According to those diagrams you will perceive less bass and treble than it was meant to be. To compensate for that we take the difference between the 70 and 80 phon curves and add it to frequency response. You now can listen at lower levels and still perceive the same tonality. That's usually called loudness.

Harman aims to emulate a flat speaker in a room, compensating for the body and ear influences that and over ear or in ear may not have

No, Harman never intended something like this. They only wanted to monitor listener preferences.
They did however use a set of speakers in their reference room measured through their artificial head as a baseline for their test. The baseline thought wasn't derived from a flat pair of speakers but a flat frequency response at the listening position, which is not the same.

 

[dongster]

No. At least I hope so.

What we're looking at are SPL levels for pure sine tones that will result in the same perceived loudness regardless of the frequency. Those graphs answer questions like: If 1 kHz plays at 80dB SPL, how loud must be 2 kHz to be recognized as equally loud?

Regarding IEM measurements we usually talk about about the frequency response at the eardrum (DRP). That's a whole nother story.

The only information we can take out of the above diagrams and use for IEM tuning is the difference between two of those curves.
Let's say your music is mastered to be heard at 80dB SPL (@1 kHz). But you want to protect your hearing and like to listen at 10dB less volume. According to those diagrams you will perceive less bass and treble than it was meant to be. To compensate for that we take the difference between the 70 and 80 phon curves and add it to frequency response. You now can listen at lower levels and still perceive the same tonality. That's usually called loudness.


No, Harman never intended something like this. They only wanted to monitor listener preferences.
They did however use a set of speakers in their reference room measured through their artificial head as a baseline for their test. The baseline thought wasn't derived from a flat pair of speakers but a flat frequency response at the listening position, which is not the same.

hi bartzky,

thanks for the response, its nice always nice to find someone who enjoys talking about such topics

i just re-read the article from innerfidelity.
https://www.innerfidelity.com/content/acoustic-basis-harman-listener-target-curve

addressing ur 2 points:

1) u are totally right, i misspoke, thats not their aim, but that's effectively what the introduction of a dummie shows
since the 3k hump and bass shelf shelf are both appear due to the introduction of dummie
2) thats what i meant with "flat speaker in a room", i did not mean free field flat

if there is some other sources that you could point me to (deutsch wäre auch passend), id love to learn more on the subject

as to your opposing stance on incorporation of hearing sensitivity into tuning:
i imagine youd agree? that it has to be incorporated somewhere, microphone design, mixing or headphone

 

[dongster]

No. At least I hope so.

What we're looking at are SPL levels for pure sine tones that will result in the same perceived loudness regardless of the frequency. Those graphs answer questions like: If 1 kHz plays at 80dB SPL, how loud must be 2 kHz to be recognized as equally loud?

Regarding IEM measurements we usually talk about about the frequency response at the eardrum (DRP). That's a whole nother story.

The only information we can take out of the above diagrams and use for IEM tuning is the difference between two of those curves.
Let's say your music is mastered to be heard at 80dB SPL (@1 kHz). But you want to protect your hearing and like to listen at 10dB less volume. According to those diagrams you will perceive less bass and treble than it was meant to be. To compensate for that we take the difference between the 70 and 80 phon curves and add it to frequency response. You now can listen at lower levels and still perceive the same tonality. That's usually called loudness.


No, Harman never intended something like this. They only wanted to monitor listener preferences.
They did however use a set of speakers in their reference room measured through their artificial head as a baseline for their test. The baseline thought wasn't derived from a flat pair of speakers but a flat frequency response at the listening position, which is not the same.

oh and do you have thoughts on target curve? having measured quite a number of iems yourself

 

[bartzky]

1) u are totally right, i misspoke, thats not their aim, but that's effectively what the introduction of a dummie shows
since the 3k hump and bass shelf shelf are both appear due to the introduction of dummie

Alright, I get your thoughts. The bass bump though is not due to the head. Take a look at the green curve from this diagram:

(Taken from the Innerfidelity article you've linked)

The green curve was the baseline for their research. As you can see the bass bump is not present there.

if there is some other sources that you could point me to (deutsch wäre auch passend), id love to learn more on the subject

I absolutely recommend AES. In their library you can find all those papers by Olive, Welti, Toole, Temme, Lorho and many others and get all the information first hand. E.g. those papers we are talking about right now:
- http://www.aes.org/e-lib/browse.cfm?elib=17042
- http://www.aes.org/e-lib/browse.cfm?elib=17940

And their newest paper that includes a new 2017 Harman curved derived for IEMs:
- http://www.aes.org/e-lib/browse.cfm?elib=18369

An AES membership is not free though...

Soon I'd like to publish a German article on my website in which I like to address target curves and stuff (viel auf Deutsch gibt's bisher ja noch nicht zu dem Thema).

If you have any questions don't be shy to ask @crinacle's thread recently became the thread-to-be for any of those questions. You also can send me PMs, if you prefer to read some German words.

Are you a member of hifi-forum.de?

as to your opposing stance on incorporation of hearing sensitivity into tuning:
i imagine youd agree? that it has to be incorporated somewhere, microphone design, mixing or headphone

Well, a common listening level for studios is at 83 dB SPL average. We have to assume that this is about the SPL the music is intended to be reproduced at. If there has been done any loudness compensation yet, it likely has been done in regard to this level. So if you like to listen at those levels there isn't any need to compensate. But if not, aiming for a loudness compensation might not be a bad idea

oh and do you have thoughts on target curve? having measured quite a number of iems yourself

Absolutely And I do have my own target that we use internally for our reviews.
May I point you to my comment at Innerfidelity? It does include my curve and some other thoughts as well:
https://www.innerfidelity.com/comment/513668#comment-513668

 

[dongster]

Alright, I get your thoughts. The bass bump though is not due to the head. Take a look at the green curve from this diagram:

(Taken from the Innerfidelity article you've linked)

The green curve was the baseline for their research. As you can see the bass bump is not present there.


I absolutely recommend AES. In their library you can find all those papers by Olive, Welti, Toole, Temme, Lorho and many others and get all the information first hand. E.g. those papers we are talking about right now:
- http://www.aes.org/e-lib/browse.cfm?elib=17042
- http://www.aes.org/e-lib/browse.cfm?elib=17940

And their newest paper that includes a new 2017 Harman curved derived for IEMs:
- http://www.aes.org/e-lib/browse.cfm?elib=18369

An AES membership is not free though...

Soon I'd like to publish a German article on my website in which I like to address target curves and stuff (viel auf Deutsch gibt's bisher ja noch nicht zu dem Thema).

If you have any questions don't be shy to ask @crinacle's thread recently became the thread-to-be for any of those questions. You also can send me PMs, if you prefer to read some German words.

Are you a member of hifi-forum.de?


Well, a common listening level for studios is at 83 dB SPL average. We have to assume that this is about the SPL the music is intended to be reproduced at. If there has been done any loudness compensation yet, it likely has been done in regard to this level. So if you like to listen at those levels there isn't any need to compensate. But if not, aiming for a loudness compensation might not be a bad idea


Absolutely And I do have my own target that we use internally for our reviews.
May I point you to my comment at Innerfidelity? It does include my curve and some other thoughts as well:
https://www.innerfidelity.com/comment/513668#comment-513668

i really appreciate your detailed response
big help to clear some of my misconceptions and see some new perspectives

checked out AES, membership is quite pricey,
but i think ill bite the bullet when I have more free time to read these, its a fascinating subject.
edit: damn they my college is not listed there, feels bad man

edit. only browsed hifi.de before, didnt register yet, ive seen quite a number of your posts

looking forward to your article~

 

[castleofargh]

i really appreciate your detailed response
big help to clear some of my misconceptions and see some new perspectives

checked out AES, membership is quite pricey,
but i think ill bite the bullet when I have more free time to read these, its a fascinating subject.
edit: damn they my college is not listed there, feels bad man

edit. only browsed hifi.de before, didnt register yet, ive seen quite a number of your posts

looking forward to your article~

I'd suggest to start here http://seanolive.blogspot.com/
while sadly many of the papers are not on free access anymore(they were for a long time), you still get the essence in the blog, some graphs and sometimes a nice little video with a slideshow. to start on the first research about headphone measurements and preferences by Harman you have to go look as far as early 2013 in the blog. they tested a lot of things with a lot of people, including different continents just to check if there were regional taste or stuff like that. that's how far they went... (bad pun is bad).
but all that is focused on headphones and personal preferences, not IEMs and not directly neutral(but it's assumed to be the preferred signature sometimes). so some added aspects can come to play with IEMs like the obvious insertion depth and choice of tips.

 

[dongster]

found a presentation done by Sean Olive himself on 5 of his research questions (see pic) in the AES papers, recommended by @bartzky :


note: as clarified earlier
the target curve based on the how a speakers sounds in a room (flat in that room), with preference adjustment from surveys

table of contents from Dr.Olives presentation:

 

[dongster]

I'd suggest to start here http://seanolive.blogspot.com/
while sadly many of the papers are not on free access anymore(they were for a long time), you still get the essence in the blog, some graphs and sometimes a nice little video with a slideshow. to start on the first research about headphone measurements and preferences by Harman you have to go look as far as early 2013 in the blog. they tested a lot of things with a lot of people, including different continents just to check if there were regional taste or stuff like that. that's how far they went... (bad pun is bad).
but all that is focused on headphones and personal preferences, not IEMs and not directly neutral(but it's assumed to be the preferred signature sometimes). so some added aspects can come to play with IEMs like the obvious insertion depth and choice of tips.

thanks for the tip

it should apply to IEMs tho, as the measurements are taken at ear drum, and the preferences are represented in the form of ear drum measurements as well, as is Harman Target itself,

edit:
in Tylls articles, you can find comments on how an IEM fits the Harman, notable performers according to Tyll incl. Titan1, GR07

 

[DanWiggins]

One thing to ALWAYS remember when looking at measurements and such - the equipment AND environment AND tester all matter. You can look for consistency between measurements within a single measurement system, but don't expect that because IEM X measures a certain way on rig Y that it will measure the same on rig Z. Even differences between 711 couplers exist (as the standard is a bit loose in terms of the resonant frequency - the primary resonance can be set anywhere from 12 kHz to 15 kHz and be within spec).

In other words, if the measurements in this thread are made consistently in terms of equipment, processing, environment (background noise can affect you) and tester (same pressure applied, same depth, etc) then the measurements made here would be relatively representative between IEM and headphone models. HOWEVER, I would not expect them to correlate dB-for-dB with a system like Jude has, with a radically different ear (the new GRAS unit), equipment (full AP system), environment (closed chamber), etc. Nor with Tyll - again, different couplers, environments, etc. Nor even with my own measurements - different couplers, environments, etc.

All that said - measurements between units can be interesting to see from a single reviewer, but please do NOT try to correlate measurements from one set to another without a LOT of ground-truthing over dozens of units and an extended amount of time and effort.

 

[dongster]

One thing to ALWAYS remember when looking at measurements and such - the equipment AND environment AND tester all matter. You can look for consistency between measurements within a single measurement system, but don't expect that because IEM X measures a certain way on rig Y that it will measure the same on rig Z. Even differences between 711 couplers exist (as the standard is a bit loose in terms of the resonant frequency - the primary resonance can be set anywhere from 12 kHz to 15 kHz and be within spec).

In other words, if the measurements in this thread are made consistently in terms of equipment, processing, environment (background noise can affect you) and tester (same pressure applied, same depth, etc) then the measurements made here would be relatively representative between IEM and headphone models. HOWEVER, I would not expect them to correlate dB-for-dB with a system like Jude has, with a radically different ear (the new GRAS unit), equipment (full AP system), environment (closed chamber), etc. Nor with Tyll - again, different couplers, environments, etc. Nor even with my own measurements - different couplers, environments, etc.

All that said - measurements between units can be interesting to see from a single reviewer, but please do NOT try to correlate measurements from one set to another without a LOT of ground-truthing over dozens of units and an extended amount of time and effort.

thanks man, nice read

any idea what could cause x-axis shifts? meaning the shape is the same, but the peaks and dips are shifted left or right in x direction.

 

[castleofargh]

thanks man, nice read

any idea what could cause x-axis shifts? meaning the shape is the same, but the peaks and dips are shifted left or right in x direction.

resonance can do that. think about blowing into a bottle, if you add water to change the volume of air inside it, you change the frequency at which the bottle resonates. by inserting the IEM deeper you change at the very least the volume of air in the ear canal/IEM system. that's only the super obvious stuff.
the measurement rig obeys the same rules, but it's not so strange that a perfect cylinder and some more or less arbitrary distance between the IEM and the mic, will end up showing something very different. maybe we just assume the spike shifted because we see/ear 2 spikes and assume they must be the same thing. but your ear is not Crinacle's coupler.
again as others have said, of course we wish to correlate measurements and our impressions, I'd like that too. but most of the time we are talking about 2 different systems and conditions and there is only so much to correlate without making false assumptions. the purpose of those measurements if to tell when one IEM has a lot more trebles than another one. measurement compared with other measurement done the same way with the same rig. then because you've heard one of the IEMs, you can start making educated guesses about how the other might sound to you. that rational is pretty solid. at least more than a direct measurement to feeling conversion attempt based on 1 graph and some arbitrary compensation.

 

[crinacle]

thanks man, nice read

any idea what could cause x-axis shifts? meaning the shape is the same, but the peaks and dips are shifted left or right in x direction.

resonance can do that. think about blowing into a bottle, if you add water to change the volume of air inside it, you change the frequency at which the bottle resonates. by inserting the IEM deeper you change at the very least the volume of air in the ear canal/IEM system. that's only the super obvious stuff.
the measurement rig obeys the same rules, but it's not so strange that a perfect cylinder and some more or less arbitrary distance between the IEM and the mic, will end up showing something very different. maybe we just assume the spike shifted because we see/ear 2 spikes and assume they must be the same thing. but your ear is not Crinacle's coupler.
again as others have said, of course we wish to correlate measurements and our impressions, I'd like that too. but most of the time we are talking about 2 different systems and conditions and there is only so much to correlate without making false assumptions. the purpose of those measurements if to tell when one IEM has a lot more trebles than another one. measurement compared with other measurement done the same way with the same rig. then because you've heard one of the IEMs, you can start making educated guesses about how the other might sound to you. that rational is pretty solid. at least more than a direct measurement to feeling conversion attempt based on 1 graph and some arbitrary compensation.

Addressed this observation quite early on in the thread here. Deeper insert results in higher frequency resonant peaks shifting higher, while a shallower insert results in a shift to a lower frequency. This is not limited to DIY tube-based couplers and even ISO-standard couplers need to take this into consideration.

 

[DanWiggins]

Precisely. I have a little jig that presses the IEMs into a fixed location relative to my couplers, and I'm able to get +/- 0.1 dB repeatability between measurements for the same unit. I also calibrate for temperature and humidity, both of which will affect measurements - it's why you use a golden sample every few hours on your production run to "recenter" the pass/fail window based on current conditions.

 

[bartzky]

checked out AES, membership is quite pricey,
but i think ill bite the bullet when I have more free time to read these, its a fascinating subject.
edit: damn they my college is not listed there, feels bad man

I can feel you. My college isn't listed as well.
If I remember correctly it was about 43€ for a one year student subscription. Well worth it IMO.

I'd suggest to start here http://seanolive.blogspot.com/

How could I forget that? Very nice blog. I hope he will continue to fill it with more content!

What just came to my mind:
- https://www.etymotic.com/publications

A lot of their papers are about hearing and audiology, but there's some nice information about diffusefield targets as well.

- http://vbn.aau.dk/en/publications/d...rs(0e8e8200-6a84-11dc-a97d-000ea68e967b).html

A nice paper from Hammershoi and Moller.

found a presentation done by Sean Olive himself on 5 of his research questions (see pic) in the AES papers, recommended by @bartzky :

Good find!

it should apply to IEMs tho, as the measurements are taken at ear drum, and the preferences are represented in the form of ear drum measurements as well, as is Harman Target itself,

edit:
in Tylls articles, you can find comments on how an IEM fits the Harman, notable performers according to Tyll incl. Titan1, GR07

In theory it should apply to IEMs as well. Harman's most current research topic is if that's really true or if listeners prefer different curves with IEMs than with headphones. Spoiler: They do prefer 5.1 dB more bass.
Still, there needs to be more research done. Currently they only tested for the preferred low frequency response with a group of ten trained listeners.

Also IEMs and headphones are measured differently. Welti stated that there are leakage effects when measuring headphones with standard ears that will cause an underrepresentation of bass of about 3 dB compared to real ear measurements. Most likely this effect isn't present in IEM measurements. So if we compare IEM and headphone measurements we should keep in mind that the headphone might have more bass than we are actually seeing in there measurement. So at least about 3 dB of those 5.1 dB might be explained by this effect.

Todd Welti's paper can be found here: http://www.aes.org/e-lib/browse.cfm?elib=17699

On a side note: I've just put my new IEC 60318-4 compliant gear into operation some days ago